PointTrackNet: An End-to-End Network For 3-D Object Detection and Tracking From Point Clouds
This addresses multi-object tracking in 3-D point clouds for applications like autonomous driving, but it is incremental as it builds on existing machine learning-based frameworks.
The paper tackles the problem of 3-D object detection and tracking from point clouds, particularly under extreme motion conditions like sudden braking and turning, by proposing PointTrackNet, an end-to-end network that generates foreground masks, 3-D bounding boxes, and tracking displacements; it achieves competitive results on the KITTI tracking dataset, especially in irregular and rapidly changing scenarios.
Recent machine learning-based multi-object tracking (MOT) frameworks are becoming popular for 3-D point clouds. Most traditional tracking approaches use filters (e.g., Kalman filter or particle filter) to predict object locations in a time sequence, however, they are vulnerable to extreme motion conditions, such as sudden braking and turning. In this letter, we propose PointTrackNet, an end-to-end 3-D object detection and tracking network, to generate foreground masks, 3-D bounding boxes, and point-wise tracking association displacements for each detected object. The network merely takes as input two adjacent point-cloud frames. Experimental results on the KITTI tracking dataset show competitive results over the state-of-the-arts, especially in the irregularly and rapidly changing scenarios.